通过监测推理中的不确定性信号，实现更高效可靠的LLM早停。
Statistical Early Stopping for Reasoning Models
- 用统计方法检测推理过程中的不确定性关键词出现间隔
- 在数学推理任务中效率提升显著，误停率可控
- 适合需要减少冗余推理的高可靠性场景
尽管大语言模型在推理能力上取得显著进步，但在不确定情况下仍可能过度思考，产生不必要的推理步骤，尤其面对表述不清或模糊的问题时。本文提出两种基于统计原理的早停方法：第一种为参数化方法，将不确定性关键词的间隔建模为更新过程，并应用序贯检验决定停止；第二种为非参数方法，在已知样本量下提供过早停止概率的严格保证。我们在多个领域和模型上进行了实证评估，结果表明，基于不确定性的早停机制可同时提升推理效率与可靠性，尤其在数学推理任务中表现突出。
原文摘要 · Abstract (English)
While LLMs have seen substantial improvement in reasoning capabilities, they also sometimes overthink, generating unnecessary reasoning steps, particularly under uncertainty, given ill-posed or ambiguous queries. We introduce statistically principled early stopping methods that monitor uncertainty signals during generation to mitigate this issue. Our first approach is parametric: it models inter-arrival times of uncertainty keywords as a renewal process and applies sequential testing for stopping. Our second approach is nonparametric and provides finite-sample guarantees on the probability of halting too early on well-posed queries. We conduct empirical evaluations on reasoning tasks across several domains and models. Our results indicate that uncertainty-aware early stopping can improve both efficiency and reliability in LLM reasoning, and we observe especially significant gains for math reasoning.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。